Seismicity and state of stress within the overriding plate of the Tonga-Kermadec subduction zone

Seismicity and state of stress within the overriding

plate of the Tonga-Kermadec subduction zone

M.-A. Bonnardot,1 M. Regnier,1 E. Ruellan,1 C. Christova,2 and E. Tric1

Received 3 September 2006; revised 22 March 2007; accepted 5 July 2007; published 19 October 2007.

[1] To reassess the main tectonic units and to quantifythe slip partitioning within the overriding plate of theTonga-Kermadec subduction zone, a seismotectonicstudy was performed using global seismicity and focalmechanisms catalogs. (1) New tectonic features wereidentified within the Lau Basin and the volcanic arc byremarkable shallow hypocenters alignments. (2) TheCentroid Moment Tensor solutions catalog wasprocessed in order to map the stress tensor variation inthe upper plate. We found the tectonic featurescharacterized by a diffuse seismicity are subjected to acomposite stress regime and they are interpreted asdiffuse immature plate boundaries controlled by the highthermal anomaly lying beneath the Lau Basin. (3) Wequantified the margin-parallel rates of motion using theazimuth of the maximum compressive stress componentcomputed within the interplate zone. The resultshighlight a major tectono-kinematic segmentationrelated to the subduction of the Louisville SeamountChain. Citation: Bonnardot, M.-A., M. Regnier, E. Ruellan,

C. Christova, and E. Tric (2007), Seismicity and state of stress

within the overriding plate of the Tonga-Kermadec subduction

zone, Tectonics, 26, TC5017, doi:10.1029/2006TC002044.

1. Introduction

[2] The Tonga-Kermadec subduction zone is part of theextended Australia-Pacific plate boundary and reflects amultistage tectonic history related to global rearrangementsof plate convergence in the SW Pacific [Hamburger andIsacks, 1987; Sdrolias et al., 2001, 2003]. Many geodynamicprocesses contribute to the complex present-day tectonicpattern observed along the 2700 km of the Tonga-Kermadecsystem [Ruellan and Lagabrielle, 2005; Pelletier et al.,1998].[3] This subduction system is characterized by a N15�E

trending back-arc domain that is mostly parallel to thevolcanic arc [Karig, 1971]. The back-arc domain exhibitsstrong variations of the state of stress and of the orientationsof the tectonic structures from north to south [Ballance et

al., 1989; Ruellan et al., 2003; Delteil et al., 2002; Parsonand Wright, 1996]. Indeed, back-arc spreading is wellestablished in the Lau Basin in contrast to back-arc riftingwithin the Havre Trough [Karig, 1971]. This structuralpattern highlights a major tectonic segmentation of thewhole system, which coincides with the subduction of theaseismic Louisville Seamount Chain (LSC). Owing to itsobliquity relative to the trench and the plate motion azimuthof the Pacific subducting plate, the LSC is sweeping themargin southward and thus it is inferred to control the Lauback-arc basin opening by a collision-induced volcanic arcrotation process [Ruellan et al., 2003; Wallace et al., 2004].Indeed, GPS observations, sea floor magnetization andbathymetry data revealed a north to south gradient of theoceanic opening rates implying a global V-shape of the Laubasin [Honza, 1995; Taylor et al., 1996; Fujiwara et al.,2001]. These spreading rates, measured from GPS observa-tions, reach up to 159 ± 10 mm/yr at 16�S and decrease to91 ± 4 mm/yr at 21�S [Bevis et al., 1995] and many authorsproposed that the spreading centers are propagating south-ward [Parson and Wright, 1996; Ruellan et al., 2003]. Onthe contrary, the uniform width of the Havre Trough fromnorth to south without any identified spreading centers mayreflect the initial stage of back-arc rifting, where the back-arc opening process would be locked [Ballance et al., 1999;Nishizawa et al., 1999; Wright, 1997].[4] The northern termination of the Tonga-Kermadec

subduction is controlled by additional mechanisms, thatare expected to affect locally the state of stress within theoverriding plate. First, the northern edge of the Tonga trenchis characterized by the tearing of the Pacific subductingplate. On the basis of the earthquake distribution and source-mechanism determinations, Millen and Hamburger [1998]showed that the Pacific plate is progressively downwarpedas it enters into the northern part of the trench and tornfrom 18 down to 88 km in depth over the entire litho-spheric thickness. Second, the trench extends westward asa large transform fault with a right lateral strike-slip motion[Eguchi, 1984], that accommodates the westward Pacificplate motion.[5] Despite the previous numerous results obtained on the

Tonga-Kermadec subduction zone, the competition betweenmajor geodynamic mechanisms precluded so far to figureout the precise evolution of the segmented back-arc basinopening. The important increase of hypocenter and focalmechanism data over the last decades provides now a largerdata set that allows to revisit the problem. Through aseismotectonics study, we attempt to reassess the present-day state of stress in the Tonga-Kermadec zone, in order toquantify the strain partitioning related to the LSC subduc-

TECTONICS, VOL. 26, TC5017, doi:10.1029/2006TC002044, 2007ClickHere

for

FullArticle

1Universite Nice-Sophia Antipolis, Institut de Recherche pour leDeveloppement (I.R.D.), Centre National de Recherche Scientifique(CNRS), Laboratoire Geosciences Azur, Valbonne, France.

2Department of Seismology, Geophysical Institute of BulgarianAcademy of Sciences, Sofia, Bulgaria.

Copyright 2007 by the American Geophysical Union.0278-7407/07/2006TC002044$12.00

TC5017 1 of 15

http://dx.doi.org/10.1029/2006TC002044

tion-induced segmentation. We finally propose an updatedtectono-kinematic model for the Tonga-Kermadec systemthat will allow to further address the question of regionalgeodynamic evolution.

2. Distribution of Shallow Seismicity and

Focal Mechanisms

[6] On the basis of the distribution of shallow seismicactivity (Figure 1), the Tonga-Kermadec subduction zone isfrequently divided into three domains corresponding to theprincipal tectonic units, i.e., the interplate zone, the arc-backarc domains and the northwestern area [Isacks et al., 1969;Louat and Dupont, 1982; Pelletier and Louat, 1989; Sykeset al., 1969].

[7] In this study, we analyzed the shallow seismicity fromthe Engdahl catalog (earthquake depth <50 km) (Figure 1a)[Engdahl et al., 1998] together with the focal mechanismssolutions (CMT) (Figure 1b) to model the shape and thestate of stress of the slab [Dziewonski et al., 1981]. Weclassified the focal mechanisms into three groups based onthe value of the P- and T-axis dip: the reverse group(T-axis dip � 45�), the normal group (P-axis dip � 45�)and the strike-slip group (P and T-axis dip < 45�). Westudied four domains with preferential types of faulting:(zone a) the interplate area, (zone b) the Kermadec-Havredomain, (zone c) the Tonga-Lau domain, and (zone d) thelarge northern part of the Lau basin (insert in Figure 1b). TheKermadec-Havre (zone b) and the Tonga-Lau (zone c)domains were studied separately owing to their present-

Figure 1. (a) Distribution of the shallow seismicity (<50 km) in both plates, from the Engdhal catalog[Engdahl et al., 1998]. The bathymetry data are from Zellmer and Taylor [2001]. The two newlyidentified structures are highlighted: the Futuna-Niua Fo’ou lineament and the intra-arc Niuatoputapulineament. NFFZ, North Fiji Fracture Zone. (b) Distribution of the CMT solutions within the upper platefrom Dziewonski et al. [1981]. The four main domains referred into the text correspond to: zone a,Interplate zone; zone b, Kermadec–Havre Trough zone; zone c, Tonga-Lau zone; and zone d, northernarea.

TC5017 BONNARDOT ET AL.: STRESS REGIME IN THE TONGA-KERMADEC AREA

2 of 15

TC5017

day different tectonic regime, that is, a back-arc rifting stageand a spreading stage, respectively.

2.1. Interplate Area

[8] This domain (zone a in Figure 1b insert) lying betweenthe modern volcanic arc zone and the trench, is the mostactive zone in the Tonga-Kermadec system. The underthrust-ing of the Pacific subducting plate explains the predominantthrust mechanisms observed along the interplate zone. Seis-mic gaps clearly underline the segmentation of the N15�Etrending interplate zone and may eventually reflect differentsegments of the seismic cycle. An alternate interpretationmay be provided by the good correlation between thepositions of these seismic gaps and of oceanic reliefs acrossthe margin [Vogt et al., 1976; Scholz and Small, 1997]. Twolarge gaps, at 25–27�S and 33–40�S, are indeed localized infront of the Louisville Seamount Chain (LSC) and the

Hikurangi Plateau. These topographic highs are expectedto increase the interplate coupling, eventually locking thesubduction and therefore, they may strongly disturb thestress regime in the upper plate. A third smaller seismicgap is roughly centered at 18�S, but it may be related to aninherited feature, since the small seamount lying in front of itis not yet subducting into the trench.[9] The seismicity distribution of the Wadati-Benioff

zone down to 200 km in depth reveals some geometryvariations of the subducting slab along the Tonga-Kermadectrench (Figure 2, left). In order to get a sharper image of theseismicity and to reassess the subducted plate geometry, weprocessed the global seismicity catalog from the Engdahldatabase [Engdahl et al., 1998] using the statistical methoddescribed by Bossu [2000]. This method assumes that, for agiven hypocenter associated with a location uncertaintyellipsoid, all the other earthquakes located inside this

Figure 2. (left) The 200-km-width cross sections into the slab of the global seismicity and (right)collapsed seismicity, computed thanks to the collapse Bossu’s method [Bossu, 2000] applied to theEngdahl catalog [Engdahl et al., 1998] and using a 60-km-diameter sphere. The fits of the Wadati-Benioff Zone were computed from 100-km-width cross sections at each latitude degree along the trenchand reflect the variation between the shallow slab dip, a (<50 km) and the deep slab dip, b (>50 km) (±5�precision). They were used to separate the CMT belonging to the upper plate from those belonging to thedowngoing plate.


3 of 15

TC5017

ellipsoid belong to the same active structure. The centroid ofall these considered hypocenters is computed and is taken asthe best estimate of the variable location. This method allowsfor simplifying the complex image produced by a largedatabase in reducing the effects of random location uncer-tainties. For simplicity we have used a sphere instead of anellipsoid without significant change on the output pattern.Intensive tests have been carried out to determine an averageoptimum diameter. The right plots in Figure 2 illustrate thecomputed collapsed seismicity when using a 60 km diametersphere. Thanks to this method, we greatly improve theseismic slab image and we are able to numerically fit by athird-order polynome curve the Wadati-Benioff zone with a5� dip uncertainty. Consequently, in Figure 2, we see that theseismic gaps are observed in zones where the slab dip at theinterplate contact is lower (a = 16–18� ± 5, Figures 2a and2c), suggesting a strong interplate coupling. On the contrary,the most active seismic zones display steeper slab dip (a =27–31� ± 5, Figures 2b and 2d).[10] At the northern termination, the tearing of the Pacific

plate is emphasized by the predominant normal focalmechanisms and induces a high level of seismic activity[Millen and Hamburger, 1998]. We note another activeseismic zone slightly oblique to the main interplate areabetween 16�S and 17�S, referred as the Niuatoputapulineament (Figure 1a), where a surprising high density ofnormal and strike-slip mechanisms is observed in thisthrust-dominated domain.

2.2. Kermadec Arc-Havre Trough

[11] This domain corresponds to an arc-back arc system(zone b in Figure 1b). Normal and strike-slip faulting areobserved on the western flank of the Kermadec arc. Thistectonic activity may be related to the major strike-slip fault,that accommodates the along-strike component of theoblique convergence and controls the rifting of the HavreTrough. No spreading center was identified in this back-arcdomain to attest to a present-day back-arc opening, but amean opening rate of 17 mm/yr was deduced from thestructural pattern [Delteil et al., 2002]. Strike-slip faultingcharacterizes the area north of 32�S, whereas normalmechanisms are mainly observed south of this latitude.Such a pattern may likely support the existence of a majortectonic boundary at 32�S [Pelletier and Dupont, 1990;Delteil et al., 2002]. The southern end of the studied areacorresponds to the Taupo volcanic zone in the North Islandof New Zealand, which is interpreted as an active back-arcbasin undergoing consequently normal faulting [Parson andWright, 1996]. This complex region is controlled by theoblique subduction of the Hikurangi plateau and farthersouth, around 44�S, by the Chatham Rise [Herzer et al.,2000; Delteil et al., 2003]. These oceanic plateaus induce astrong interplate coupling between the two plates asevidenced by the large seismic gap. Note that this seismicgap extends northward of the Hikurangi plateau, whosenorthern limit ends at 36�S. The gap would therefore reflectan inherited feature that Collot and Davy [1998] associatedwith the initial collision point between the Hikurangiplateau and the Kermadec trench.

2.3. Tonga Arc-Lau Basin

[12] This unit extends roughly from 16�S to 25�S (zone cin Figure 1b) and depicts a very clustered seismicity, whichhighlights the main tectonic features. A clear seismicalignment underlines the Peggy ridge. On the contrary, amore diffuse group of strike-slip mechanisms is likelycoincident with the central overlapping spreading centersof the Lau basin showing a left stepping geometry [Zellmerand Taylor, 2001]. Both the Valu Fa spreading ridge (endingat 24�S) and North East Lau Spreading Center (ending at18�S) mostly exhibit seismic activity at their southerntermination. These high concentrations of normal mecha-nisms is in good agreement with the southward propagationof these volcanic ridges.

2.4. Northern Area

[13] As previously mentioned, the complexity of thenorthern termination imposes to treat this region separately.It corresponds to the eastern end of the North Fiji FractureZone (NFFZ) and extends from 14�S to 16�S (Figure 1b). Itis characterized by numerous strike-slip mechanisms scat-tered in one degree-wide strip stretching along the southernside of the plate boundary. A small group of reverse faultsnorth of the Futuna Island indicates a local zone ofcompression along the NFFZ. They were interpreted byRegnier [1994] as compressional relay zones along themajor strike-slip North Futuna Transform Zone (NFTZ).The distribution of the shallow earthquakes reveals aseismic alignment extending from the Futuna Island to theNiua Fo’ou Island that is coincident with a poorly knowntopographic high (the Rochambeau Bank). It lengthenssouthward the NFTZ described by Regnier [1994] andPelletier et al. [2001]. We note this lineament has anintermediate orientation between the azimuth of the Peggyridge and of the northern termination of the Tonga trench.

3. Inversion Method for Resolution of Stress

Tensors

[14] To perform our inversion study, we first selected thefocal mechanisms solutions from the Harvard catalog on thebasis of the three statistical criteria described by Frohlichand Apperson [1992] and Frohlich and Davis [1999], i.e.,the relative error Erel (erel < 0.30), the non-double-couplecomponent (fclvd < 0.25) and the degree of freedomintroduced in the inversion (nfree = 6). This selection allowsus for using the better-constrained moment tensors thenreducing the uncertainties in the stress inversion computa-tion. Applying these criteria, about 20% of the CMTsolutions were eliminated. Then we selected the eventslocated above the computed fit of the Wadati Benioffseismic zone presented in Figure 2, assuming they allbelong to the upper plate. To the detriment of precision,this method has the benefit to select and analyze a data setin an objective way.[15] We computed the stress regime by inverting the

P- and T-axis data for the CMT located within each selectedarea using the Gephart’s inversion method (FMSI) [Gephart


4 of 15

TC5017

and Forsyth, 1984; Gephart, 1990]. This method assumesthat: (1) the stress tensor is uniform in the space and timeconsidered; (2) earthquakes are shear dislocations and theycan occur on preexisting faults; and (3) the differencebetween the direction of the computed shear stress on thefault plane and the observed slip on a fault plane is

minimized. FMSI determines the directions of the threestress tensor components (s1 > s2 > s3) and the stress shaperatio defined as R = (s2 � s1)/(s3 � s1), with 0 � R � 1.When R approaches 0, s1 ’ s2, that means a biaxialdeviatoric compression; when R approaches 1, the switchbetween s3 and s2 indicates an uniaxial compression

Figure 3. Map showing the P- and T-axis orientations for each selected small area in the Tonga zone.The great circle on the stereograms corresponds to the interplate plane. The two major 05/03/2006 eventsare localized on the map with open and solid stars. NFFZ, North Fiji Fracture Zone; FSC, FutunaSpreading Center; PR, Peggy Ridge; FFZ, Futuna Fracture Zone; NELSC, Northeast Lau SpreadingCenter; NWLSC, Northwest Lau Spreading Center; CLSC, Central Lau Spreading Center; ELSC, EasternLau Spreading Center; VFR, Valu Fa Ridge; SLR, Southern Lau Rift.


5 of 15

TC5017

[Guiraud et al., 1989]. The best stress model is this onewhich minimizes the misfit between observed and comput-ed slip on the fault plane. The quality of the inversion isevaluated through the estimation of the 95% confidencelimits of the best-fit model [Gephart and Forsyth, 1984].Besides, assuming that initial errors are introduced in the P-and T-axes determination [Frohlich and Davis, 1999], weconsidered that the stress directions are well constrainedwhen the misfit angle, q, is q < 6� and the 95% confidencelimits of s1 and s3 do not overlap. Some heterogeneities inthe stress field are suggested for 6� < q < 10� and 10� < qrepresents a heterogeneous stress regime [Gillard et al.,1996; Lu et al., 1997].[16] In regard to the main tectonic features, special

attention was carried on the separation between areasdefined by diffuse deformation and those characterized bya pattern of identified regional faults. Thus we carefullyindividualized small areas where an homogeneous state ofstress was expected. Figures 3 and 4 illustrate the orienta-tions of the P and T axis for each of the subdivision in theTonga and Kermadec zones. Since some of the areas were

poorly constrained owing to a small number of earthquakes,in several instances we merged two neighboring regionstogether to reach a sufficient number of earthquakes tosuccessfully run the inversion program. The results of stressinversions are presented in Figures 5 and 6 as well as in theTables 1 and 2.

4. Results

4.1. Arc-Back Arc Domain

[17] The stereograms 2 and 3 in Figure 5 indicate thenorthwestern part is under a well-established left-lateralstrike-slip regime with a N45�E compression that may bemostly accommodated along the main Futuna FractureZone (FFZ) and the newly identified Futuna-Niua Fo’oulineament. Considering this left-lateral strike-slip regime,the expected orientation of tensional structures is in goodagreement with the N45�E displayed spreading centers inthe northern part of the Lau Basin. It is worthwhile tonote that the stress tensor inversion (Figure 5, stereogram3) was able to resolved the bimodal state of stress in the

Figure 4. Map showing the P- and T-axis orientations for each selected small area in the Kermadeczone. The great circle on the stereograms corresponds to the interplate plane.


6 of 15

TC5017

large fracture zone along the plate boundary. Indeed, inaddition to the main strike-slip structures, some compres-sional relay zones were observed for instance, in thevicinity of the Futuna island [Regnier, 1994; Pelletier etal., 2001] and Figure 3, stereogram 1 shows that P axesfrom both the thrust events and the strike-slip eventswere fully compatible.

[18] A fundamental change in orientation of principalstress axes is observed between the FFZ and the Central LauBasin [Hamburger and Isacks, 1988; Pelletier and Louat,1989]. We resolved indeed a strike-slip regime with a 45�counterclockwise rotation of the principal stress compo-nents in the northern part of the Central Lau Basin (Figure 5,stereograms 1 and 12), and a variation of the stress field

Figure 5. Stress tensors resolution for the Tonga segment. Stereographic projections are in lowerhemisphere and correspond to the shaded and hatched areas, in the back-arc basin and in the interplatezone, respectively. They illustrate the best fit models with their 95% and 68% confidence limits of themaximum and the minimum compressive stress component, s1 (solid square) and s3 (solid circle),respectively. The misfit angle is provided for each confidence limit to estimate the homogeneity of theresolved stress regime. The histogram represents the distribution of the tested models with theirconfidence region relative to the R ratio. Results of the best fit models (in black) are presented in Table 1.


7 of 15

TC5017

toward the south that becomes purely extensional (Figure 5,stereograms 1 and 11). This composite regime is consistentwith evidences of left-stepping offsets of the spreadingcenters observed in the Central Lau Basin [Taylor et al.,

1996]. It indicates a variation in the boundary trend of theLau microplate and reflects the southward propagation ofthe spreading centers within the back-arc basin. In theeastern part of the basin, we resolved a strike-slip regime

Figure 6. Stress tensors resolution for the Kermadec segment. See Figure 5 for more precisions. Resultsof the best fit models are presented in Table 2.

Table 1. Stress Tensors in the Tonga Regiona

Area Numbers and Names N s1 Azimuth/Dip s2 Azimuth/Dip s3 Azimuth/Dip R Q, deg

1 NELSC 12 190/75 16/15 286/1 0.5 3.52 Fiji Fracture Zone 51 217/14 66/74 309/7 0.4 5.13 North Fracture Zone 56 235/5 351/79 145/10 0.4 6.04 Tear zone 75 283/7 33/66 190/23 0.8 5.25 Curved Interplate zone 29 87/46 356/0 266/44 0.4 4.76 North Interplate zone 74 105/15 14/4 270/74 0.5 4.47 Northern Tonga Arc 18 38/2 130/41 306/49 0.7 7.68 Center Interplate zone 207 115/44 245/33 355/28 0.9 4.19 Tonga Arc 47 80/9 170/4 286/80 0.2 6.410 Tonga Arc - LSC 35 172/8 66/62 266/26 0.1 4.811 Southern Lau 14 25/46 178/40 280/14 0.2 2.312 Center Lau 54 170/0 73/87 260/3 0.4 3.6

aN is number of focal mechanisms used in the inversion; s1, s2, and s3 are maximum, intermediate, and minimum compressive stress, respectively; R =(s2 � s1)/(s3 � s1) is magnitude of the stress shape ratio; q is average misfit angle between the predicted model and the observations; NELSC is NorthEast Spreading Center; and LSC is Louisville Seamount Chain.


8 of 15

TC5017

along the North East Lau Spreading Center (NELSC)(stereogram 1, Figure 5). The pattern of the 95% confidencelimits and the misfit angle value (Table 1) indicate someheterogeneities in the stress field that may also reflect thevariation of the boundary trend.[19] In summary, the FFZ, the Peggy ridge/Central Lau

Spreading Center (CLSC) and the NELSC are characterizedby bimodal stress regimes as expected from the varioustypes of focal mechanisms. These results are in agreementwith a model of diffuse rather than narrow plate boundary.This is also supported by the spread of epicenter and focalmechanism data on an average one degree wide stripcentered on the plate boundaries of the northern part, exceptto the Peggy Ridge (Figure 1a). Such a tectonic context maybe triggered by the anomalous high thermal anomaly lyingbeneath the North Fiji Basin and Lau Basin [Lagabrielle etal., 1997], that would induce intraplate deformation as wellas plate boundary deformation. Therefore, given theseobservations, we suggest that the FFZ, the Peggy Ridge/Central Lau Spreading Center (CLSC) and the NELSCcorrespond to diffuse immature plate boundaries, interme-diate between a model of intraplate deformation without anyplate boundary, as suggested by Hamburger and Isacks[1988] and a model of rigid microplates, as proposed byZellmer and Taylor [2001].[20] The Tonga arc (from 18�S to 24�S) (Figure 5,

stereogram 9), is located between the East Lau SpreadingCenter (ELSC) and the trench, and is subjected to a E–Wcompressive regime. The orientation of the principal stresscomponents in regard to the volcanic arc trend suggests thata dextral strike-slip component is also accommodated alongthe arc.[21] Compared to the Tonga segment, the Kermadec arc

(Figure 6, stereogram 1) shows a homogeneous extensionalstress field with a dextral strike-slip component. The localresolution of the stress tensor reveals slight variations indirection from north to south (Figure 6). These along-strikevariations are consistent with the structural pattern of theHavre Trough, which displays oblique to basin axis ten-sional structures, that become progressively almost parallelto the arc when approaching the 32�S boundary [Delteil etal., 2002; Ballance et al., 1999]. Unfortunately, the smallnumber of available CMT solutions in these particular areasdoes not allow us to resolve the spatial azimuthal variationsof the stress tensor at the proposed 32�S boundary.[22] The segmentation of the overall arc-back arc domain

appears to be highly controlled by the subduction of the

Louisville Seamount Chain (LSC). Indeed, in the Louisvillearea the compressive stress component (s1) becomes almostparallel to the Louisville ridge in front of the collisionpoint (Figure 5, stereogram 10) and the very low computedR value (0.1 in Table 1) supports a biaxial compression,inferred by the main Pacific plate convergence together withthe oblique Louisville oblique indentor. Besides, the upperplate area located in front of the Louisville ridge correspondsto an uplifted transition zone between the Tonga and theKermadec segments and the morpho-structural character-istics suggest that the Lau back-arc spreading ends in thisarea [Parson and Wright, 1996; Ruellan et al., 2003]. To thesouth of this domain, a strike-slip regime is instead observedin the back-arc zone (Figure 6, stereogram 1). In the sameway, we expect such a tectonic regime variation in the arealocalized in front of the subducting Hikurangi plateau. Thiszone is indeed the locus of a major seismic gap in theinterplate domain (Figure 1a), but as the main collision pointis localized further south in front of the Chatham Rise(43�S), the low stress field variation observed within thesouthern part of the Havre Trough cannot be relatedwith certainty to the Hikurangi plateau collision (Figure 6,stereogram 2). Moreover, many other major tectonic mech-anisms in this area, like oblique convergence and continentallithosphere rifting, may overprint the collision-induced tec-tonic signature and do not permit to observe a similar stressfield pattern as in the vicinity of the LSC.

4.2. Interplate Domain

[23] The stress inversion results for the interplate zone(Figure 5, stereogram 8, and Figure 6, stereograms 3 to 6)exhibit a clear downdip extension regime with a well-constrained slab-normal s1 and a down dip s3. The patternof the 95% confidence limits suggests some possibledeviations from the downdip direction and the high R value(Tables 1 and 2) underlines a clear uniaxial compressionthat can be related to the underthrusting of the oceanicPacific plate.[24] In the Tonga interplate zone (from 18 to 24�S)

(Figure 3, stereogram 9), the P- and T-axes distributiondisplays two slightly different tendencies, i.e., (pattern a)steeply plunging T axes and (pattern b) T axes distributedwithin the interplate plane and associated P axes clusteringaround the slab normal. Pattern b reveals a downdipextensional regime related to the bending of the plate andit may be accommodated along intraslab normal faults.Pattern a would correspond to the underthrusting of the

Table 2. Stress Tensors in the Kermadec Regiona

Area Numbers and Names N s1 Azimuth/Dip s2 Azimuth/Dip s3 Azimuth/Dip R Q-Misfit, deg

1 Kermadec Arc 36 160/88 42/1 312/1 0.5 5.12 Hikurangi area 44 106/79 214/3 305/10 0.5 6.03 Interplate zone A 59 93/49 337/20 233/33 0.9 3.44 Interplate zone B 109 102/54 194/1 285/36 0.5 3.35 Interplate zone C 126 105/30 12/6 319/59 0.7 3.06 Interplate zone D 100 120/38 217/10 319/50 0.9 2.3

aSee Table 1 for details.


9 of 15

TC5017

subducting plate that can induce either the creation ofinterplate thrusts faults or the reactivation in depth of hingefaults. On the basis of this observation, we notice thatthe 7.9 Mw Tonga event, the 3 May 2006, followed bythe 6.0 Mw event, are respectively consistent with these twodifferent trends (Figure 3, stereogram 9). The interplate areanorth to 18�S is discussed in the next section.

4.3. Northern Termination Domain

[25] This domain corresponds to the region to the northof 18�S that is bounded by the trench to the east and bythe North East Lau Spreading Center to the west. Themisfit values and the 95% confidence limits of our stressinversion presented in stereograms 4, 5, 6 and 7 (Figure 5)highlight some heterogeneities in the stress field. Suchresults were expected from the initial P- and T-axes distri-bution (Figure 3, stereograms 4, 5 and 7), owing to theevident mixture between the P and T axes, especiallyobserved near the termination of the trench and in the arczone. This heterogeneous pattern reflects the superimposingof multiple tectonic mechanisms that are related to thecurved shape of the trench and affect both the upper anddowngoing plates. To better resolve the stress field in thisparticular domain, we aimed at identifying the majortectonic mechanisms in separating the different trends

observed in the global distribution of the P and T axes(Figure 3, stereograms 4, 5, 6 and 7).[26] We isolated three major groups (Figure 7), based on

the global distribution of P and T axes within the northernareas. The first one is associated with the underthrusting ofthe Pacific plate, since the azimuths of the P and T axesare parallel to the direction of convergence between theAustralia and Pacific plates (Figure 7, stereograms a, b, c,and d). The second one is represented by a cluster of T axeslying within the slab dip and the P axes clustering aroundthe slab normal. It illustrates a downdip extension that canbe related to the bending and tearing of the downgoing plate(Figure 7, stereograms a, b, and c). We note a significantdifferent behavior between these two groups, since the trendof the P and T axes from the ‘‘convergence’’ group remainsunchanged along the trench despite the variation of thecurvature. Thus these oblique-slip events may induce animportant slip partitioning within the upper plate, since theycontrol the stress field (Figure 5, stereograms 5 and 6). Onthe contrary, the axes from the ‘‘bending’’ group rotatecounterclockwise from north to south following the trenchnormal and are responsible for the stress field at thenorthern hatched area termination, which is affected byplate-tearing mechanism (Figure 5, stereogram 4) [Millenand Hamburger, 1998].

Figure 7. Distribution of the shallow seismicity with the stereographic projections of the P and T axesfor the northern curved area. Three tectonic regimes are identified: the main tectonic regime induced bythe plate convergence, the downdip extension related to the plate-tearing process, and the arccompression similar to that observed in the main Tonga arc. The great circle on the stereogramscorresponds to the interplate plane. Bathymetry is from Zellmer and Taylor [2001].


10 of 15

TC5017

[27] Compared to the interplate domain farther south(Figure 1b), we note that a higher density of normal eventsare recorded at this northern termination and particularly tothe north of the inferred Niuatoputapu lineament. Besides,this tectonic structure is also emphasizes by the variation ofthe volcanic arc trend to the north of 16.5�S (Figure 7).Thus we infer that this structure plays a major role in thestrain partitioning implied by the trench curvature and itwould likely correspond to a transition zone between a‘‘normal’’ dipping southern slab and a northern tearing slab,that could laterally affect the slab up to 200 km from thetearing zone.[28] The third identified stress group characterizes the

arc region with a compressive stress regime (zone d inFigure 7), with vertical T axes and horizontal P axes. Wenote that the P and T axes show a counterclockwise rotationcompared to those within the Tonga arc farther south(Figure 3, stereogram 10). The P axes are not well clusteredand indicate some residual heterogeneities in the stressregime, which may be attributed to the competition betweenmechanisms related to the curvature of the eastern andnorthern convergent boundaries as well as to the westernextensional spreading center.

5. Discussion

[29] The stress tensor results have evidenced a consider-able tectonic segmentation of the whole system, that ismainly controlled by the LSC subduction. On the basis ofthese results, we attempted to quantify the slip partitioningwithin the upper plate and we determined the margin-

parallel rates of motion within the arc along the subductionzone. Using the theoretical model of shear on a dippingplane [Bott, 1959], Angelier et al. [1982] showed that alonga subduction plate boundary the direction of the maincompressive component, s1, was consistent with the slipdirection between the two plates inferred from the focalmechanisms. This assumption is indeed consistent with thegood correlation observed between the focal mechanismdata and our stress tensors results, that is illustrated inFigure 8. Figure 8 shows also a deviation between the slipvector azimuths and the convergence azimuths that indicatesslip partitioning occurs all along the margin.[30] In order to quantify this slip partitioning, we con-

structed the velocity vector diagram proposed by McCaffrey[1992] for oblique convergence (Figure 9a), to compute thetrench parallel component of motion inferred from thedeviation between the slip vector and the convergencevector. We used the predicted plate velocity motion fromthe NUVEL-1 model for the Kermadec segment as repre-sentative of the Pacific plate motion vector with respect tothe Australia plate [DeMets et al., 1990]. In this computa-tion, the Havre Trough opening rate was neglected owing toits small magnitude and the uncertainty of its present-dayback-arc opening. However, if we compare the directionnormal to the extensional structures within the HavreTrough to the convergence direction of the Pacific Plate[Delteil et al., 2003], the consideration of the mean inferredHavre Trough opening rate of 17 mm/yr would tend todecrease the margin-parallel motion, suggesting that the slippartitioning would be mostly accommodated within theback-arc domain. In the Tonga segment, the margin-parallelmotion of the Tonga arc block was estimated using the GPSplate velocity vectors between the Pacific and the Tongaarc plates [Bevis et al., 1995] and thus the Lau Basinopening is taken into account in the computation of the slippartitioning.[31] The results presented in Figure 9b show a clear

kinematics segmentation from north to south and we alsoobserve that the margin-parallel motion is not uniform alongthe margin, suggesting a minor segmentation within botharcs. A southward motion of the volcanic arc blocks isresolved, except in the northern end of the Tonga arc, northof the Niuatoputapu lineament at 16�S, where a northwarddisplacement is determined. These results suggest a possibleaccommodation of the slip partitioning by diffuse deforma-tion within the arcs. Thus we infer a kinematics segmenta-tion of the margin at two scales: (1) At a global scale, thevelocity segmentation coincides with the tectonic patterndiscussed previously, since it separates the Tonga segment,which is characterized by rates of motion greater than20 mm/y from the Kermadec segment, mostly characterizedby much lower rates of motion close to 15 mm/yr onaverage (Figure 9b). Indeed greater velocities are definedin the Tonga segment with a strong and abrupt velocitydecrease (from 22 mm/y to 3 mm/yr) coincident with thecollision point of the Louisville Seamount Chain. A velocitydecrease is also recorded along the Kermadec arc whenapproaching the coupled Hikurangi plateau–Chatham Rise(from 28 mm/y to 14 mm/yr). (2) At a smaller scale, we

Figure 8. Comparison between the azimuths of the slipvectors deduced from our computed s1 (stars) versus theslip vectors determined from the focal mechanisms (solidcircles) (modified from Yu et al. [1993]). The azimuths ofthe slip vector for each shallow thrust event are obtained byrotating the slip vector about the strike of focal plane intothe horizontal. Predicted plate motion directions fromNUVEL-1 model and the uncertainties are represented bydash-dotted line and dashed lines, respectively [DeMets etal., 1990].


11 of 15

TC5017

notice that the margin-parallel motion changes along-strike(Figure 9b) and even if these variations are spatially poorlyconstrained, they seem to be correlated with tectonic fea-tures (Figure 10). In the Kermadec segment, for instance, atectonic discontinuity at around 32�S is underlined by aslight velocity break and may separates the northern part ofthe Kermadec arc with a slow motion (8–12–10 mm/yrfrom North to South) to the southern part with a more rapidmotion (28–15–14 mm/yr from north to south). In theTonga segment, a strong change between a northwarddisplacement in the northern Tonga arc and a southwardmotion of the main Tonga segment occurs roughly between16 and 17�S, that is close to the Niuatoputapu lineament.[32] Two mechanisms may account for accommodating

such velocity variations along the arc: (1) a strike-slip motionalong a fault located within the weakest zone, that is, withinthe volcanic arc, or (2) a collision-induced arc rotation asproposed by Calmant et al. [2003] andWallace et al. [2005],inducing clockwise rotation of the arc block around a rotationpole, which is localized at the collision point between abathymetric high and the trench. Here the Louisville Sea-mount Chain and the Chatham Rise–Hikurangi plateauwould act as poles of rotation for the Tonga segment andthe Kermadec segment, respectively.[33] The southward migration of the Hikurangi plateau

along the trench, the large seismic gap extending to theNorth of this bathymetric high and the variation of the slabgeometry at depth on each side of the 32�S boundary[Pelletier and Dupont, 1990], led Davy and Collot [2000]

to propose that this 32�S boundary could correspond to theinitial point of collision of the Hikurangi plateau.[34] Such a tectonic pattern is not so easily identified in

the Tonga segment owing to the particular configuration tothe north, where the curved termination strongly disturbs thesystem and implies a superimposition of two tectonicsignatures: the initial collision zone of the LSC and theactive subducting plate tearing. However, in comparison tothe Kermadec segment, we propose that the Niuatoputapulineament and the aseismic intra-arc discontinuity lyingbetween 18 and 19�S, which is referred to the Fonualeidiscontinuity (Figure 10), could also be related to the initialcollision zone of the Louisville Seamount Chain [Bonnardotet al., 2006]. Besides, we showed that the Niuatoputapulineament is reactivated by the complex slab behavior at thenorthern curved termination of the trench. We infer that theNiuatoputapu structure separates the northern tearing slabexpected to apply a strong pressure on the overriding platedue to a slight slab dip, from a region with a well-established normal dipping slab where the interplate pres-sure is decreasing southward. Consequently, the lineamentwould be expected to separate an uplifted arc to the northfrom a subsident arc to the south. From bathymetric data,the northern volcanic arc, which is slightly shifted westwardand exhibits a topographic high in its forearc domain, is ingood agreement with such an interpretation (Figure 7,stereogram b). Additional data, such as vertical displace-ment measurements from GPS, would help to constrain theongoing mechanisms.

Figure 9. (a) Diagram and formula illustrating the relationship between the slip vector (SV), thepredicted plate motion velocity (PMV) and the rates of motion accommodated within the margin (SSV =Strike-Slip Vector) [Yu et al., 1993]. (b) Bar chart showing the variation of the margin-parallel rate ofmotion (SSV) from 15�S to 40�S. This figure shows a major kinematics segmentation controlled by thesubduction of the Louisville Seamount Chain (LSC), and two minor segmentations within the Tonga andthe Kermadec arcs. A positive rate of motion indicates a southward displacement and a negative onestands for a northward displacement. Error bars were computed on the basis of the s error made on theazimuth of the main compressive stress component s1.


12 of 15

TC5017

[35] We propose in Figure 10 a global kinematic model,which synthesizes both the focal mechanism data and themain tectonic structures. The northwesternmost part of thestudy area is driven by an homogeneous sinistral strike-slipregime and is considered as one single plate only, but thepresence of major faults and spreading ridges should implya subdivision of this domain into many microplates. Theidentification of the plate boundaries as present-day orinherited structures, has a fundamental importance, if weattempt to explain the evolution of the Lau Basin opening

[Pelletier and Louat, 1989; Zellmer and Taylor, 2001;Bonnardot et al., 2006].

6. Conclusions

[36] New constraints on the present-day tectonic patternin the Tonga-Kermadec region were obtained from anupdated review of the seismic activity, since the importantincrease of hypocenters and focal mechanisms data over thelast decades allowed us to refine the main tectonic units, as

Figure 10. Summary tectonic model for the Tonga-Kermadec subduction zone. Pacific-Australia plateconvergence direction are from DeMets et al. [1990]. Heavy black arrows, kinematics observed alongtectonic structures; white arrows, kinematics deduced from this study; thin black arrows, averageinterplate slip vectors deduced from the stress inversion.


13 of 15

TC5017

well as the stress regimes previously inferred by the P- andT-axis orientations [Pelletier and Louat, 1989; Pelletier etal., 1998].[37] Most of the resolved stress tensors indicate a com-

bination of stress regimes, such as the main tectonic regimerelated to the subduction process and some local stressperturbations. Owing to the characteristics of their compos-ite stress regime, we interpreted the FFZ, the Peggy Ridge,the CLSC and the NELSC as immature plate boundaries ina region with a high thermal gradient [Lagabrielle et al.,1997], that are characterized by a diffuse distribution of theshallow seismicity related to both intraplate and plateboundary deformation. The stress inversion results alsoreflect a major tectonic segmentation of the upper plateand allow to separate the Tonga domain from the Kermadecdomain in the vicinity of the entry point into the trench ofthe Louisville Seamount Chain. The quantification of theslip partitioning, which was computed using the stresstensors resolved in the interplate zone, also emphasizes asimilar kinematics segmentation. Consequently, we con-clude that subduction of oceanic bathymetric highs mayclearly control the upper plate state of stress and that theyhave to be considered as major structures in global evolu-tion of a subducting system. Such a conclusion was alsosupported by 3-D numerical experiments, that showedstrong variations of the state and stress and the topographyof the overriding plate, depending on the geometry of thesubducting oceanic ridge [Bonnardot, 2006].[38] In addition, along-strike variations of the rate of

margin-parallel motion were reported within both domains,since we resolved a nonuniform southward migration of the

whole Kermadec arc and most of the Tonga arc, as well as anorthward migration of the northernmost part of the Tongaarc. We correlated these local variations with tectonic fea-tures and we infer that a second-order segmentation contrib-utes to a diffuse deformation of the whole volcanic arc.[39] Finally, thanks to a fine analysis of the distribution of

the shallow seismicity, we also improved the image we haveof the structural pattern of the Tonga domain and weidentified two new tectonic features: (1) the Niuatoputapulineament between 16 and 17�S; we showed that it allowsfor accommodating the variations of the slab behaviorinduced by the curvature of the trench and that it controlsthe arc deformation; and (2) the Futuna-Niua Fo’ou linea-ment, which corresponds to a topographic high and is linkednorthward to the North Futuna Transform Zone (NFTZ).This structure lets us refine the structural pattern of the FijiFracture Zone (FFZ) that mainly accommodates the LauBasin opening. These newly identified structures are alsoinferred to have played a major role in the first stages of theLau Basin opening [Bonnardot et al., 2006], and wouldhave to be taken into account in the further plate tectonicsreconstructions.

[40] Acknowledgments. We are very grateful to Michael Hamburgerfor helpful discussion and an anonymous referee for constructive commentswhich helped us to improve this manuscript. The authors also thank JohnGephart for providing the last version of the stress inversion program. TheEgide SSHN financial support program ensured the Cenka Christova’s workin the Laboratoire Geosciences Azur, Universite Nice-Sophia Antipolis,France, in 2005. This is a Geosciences Azur contribution.


14 of 15

TC5017

ReferencesAngelier, J., E. Barrier, and P. Huchon (1982), Stress

trajectories and relative motion at consumingboundaries: The Hellenic subduction (Greece) andthe Philippine-Eurasia collision (Taiwan and Japan)as examples, C. R. Hebd. Seances. Acad. Sci., 294,745–748.

Ballance, P., D. W. Scholl, T. L. Vallier, and R. Herzer(1989), Subduction of a late Cretaceous seamountof the Louisville Ridge at the Tonga trench: A mod-el of normal and accelerated tectonic erosion, Tec-tonics, 8(5), 953–962.

Ballance, P., H. Follas, G. Gibson, A. Ablaev, I. Pushchin,S. Pletnev, M. Birylina, and T. Itaya (1999),Morphol-ogy and history of the Kermadec trench-arc-backarcbasin-remnant arc system at 30 to 32�S: Geophysicalprofile, microfossil and K-Ar data, Mar. Geol., 159,35–62.

Bevis, M., et al. (1995), Geodetic observations of veryrapid convergence and back-arc extension at theTonga arc, Nature, 374, 249 –251.

Bonnardot, M.-A. (2006), Etude geodynamique de lazone de subduction Tonga-Kermadec par une ap-proche couplee de modelisation numerique 3D etde sismotectonique/Geodynamical study of theTonga-Kermadec subduction zone using both 3-Dnumerical modelling and seismotectonics ap-proaches, Ph.D. thesis, 254 pp. Univ. Nice-SophiaAntipolis, Nice, France.

Bonnardot, M.-A., M. Regnier, E. Ruellan, C. Christova,and E. Tric (2006), Stress field in the Tonga BenioffZone and geodynamicalmodel for the first stages of theLau Basin opening, Eos Trans. AGU, 87(52), FallMeet. Suppl., Abstract T51C-1543.

Bossu, R. (2000), A simple approach to constrain theposition and the geometry of the seismogenic struc-tures: application to the Karthala volcano (Grande

Comores Island, Mozambique Channel), J. Seis-mol., 4, 41 –48.

Bott, M. H. (1959), The mechanics of oblique slipfaulting, Geol. Mag., 96(2), 109–117.

Calmant, S., B. Pelletier, P. Lebellegard, M. Bevis, F. W.Taylor, and D. A. Phillips (2003), New insights onthe tectonics along the New Hebrides subductionzone based on GPS results, J. Geophys. Res.,108(B6), 2319, doi:10.1029/2001JB000644.

Collot, J.-Y., and B. Davy (1998), Forearc structuresand tectonic regimes at the oblique subduction zonebetween the Hikurangi Plateau and the SouthernKermadec margin, J. Geophys. Res., 103(B1),623–650.

Davy, B., and J.-Y. Collot (2000), The Rapuhia Scarp(northern Hikurangi Plateau)—Its nature and sub-duction effects on the Kermadec Trench, Tectono-physics, 328, 269–295.

Delteil, J., E. Ruellan, I. Wright, and T. Matsumoto(2002), Structure and structural development ofthe Havre Trough (SW Pacific), J. Geophys. Res.,107(B7), 2143, doi:10.1029/2001JB000494.

Delteil, J., J.-F. Stephan, B. Mercier de Lepinay, andE. Ruellan (2003), Wrench tectonics flip at obliquesubduction: A model from New Zealand, C. R.Acad. Sci. Geosci., 335, 743–750.

DeMets, C., A. D. F. R. G. Gordon, and S. Stein (1990),Current plate motions, Geophys. J. Int., 101, 425–478.

Dziewonski, A. M., T. Chou, and J. H. Woodhouse(1981), Determination of earthquake source para-meters from waveform data for studies of globaland regional seismicity, J. Geophys. Res., 86,2825–2852.

Eguchi, T. (1984), Seismotectonics of the Fiji Plateauand Lau Basin, Tectonophysics, 102, 17–32.

Engdahl, E., R. Van der Hilst, and R. Buland (1998),Global teleseismic earthquake relocation with im-proved travel times and procedures for depth de-termination, Bull. Seismol. Soc. Am., 88, 722 –743.

Frohlich, C., and K. Apperson (1992), Earthquake focalmechanisms, moment tensors, and the consistencyof seismic activity near plate boundaries, Tectonics,11(2), 279–296.

Frohlich, C., and S. Davis (1999), How well con-strained are the well-constrained T, B, and P axesin moment tensors catalogs?, J. Geophys. Res., 104,4901–4910.

Fujiwara, T., T. Yamazaki, and M. Joshima (2001),Bathymetry and magnetic anomalies in the HavreTrough and Southern Lau Basin: From rifting tospreading in back-arc basins, Earth Planet. Sci.Lett., 185, 253–264.

Gephart, J. W. (1990), Stress and the direction of slipon fault planes, Tectonics, 9(4), 845–858.

Gephart, J. W., and D. Forsyth (1984), An improvedmethod for determining the regional stress tensorusing earthquake focal mechanism data: Applicationto the San Fernando earthquake sequence, J. Geo-phys. Res., 89, 9305–9320.

Gillard, D., M. Wyss, and P. Okubo (1996), Type offaulting and orientation of stress and strain as afunction of space and time in Kilauea’s south flank,Hawaii, J. Geophys. Res., 101, 16,025–16,042.

Guiraud, M., O. Laborde, and H. Philip (1989), Char-acterization of various types of deformation andtheir deviatoric stress tensors using microfault ana-lysis, Tectonophysics, 170, 289–316.

Hamburger, M., and B. Isacks (1987), Deep earth-quakes in the Southwest Pacific: A tectonic inter-pretation, J. Geophys. Res., 92, 13,841–13,854.


15 of 15

TC5017

Hamburger, M., and B. Isacks (1988), Diffuse back-arcdeformation in the southwestern Pacific, Nature,332, 599–604.

Herzer, R., B. Davy, A. Duxfield, J. Mascle, E. Ruellan,N. Mortimer, and C. Laporte (2000), New con-straints on the New Zealand-South Fiji Basin con-tinent-back-arc margin, C. R. Acad. Sci., Ser. IIAEarth Planet. Sci., 330, 701–708.

Honza, E. (1995), Spreading mode of backarc basins inthe western Pacific, Tectonophysics, 251, 139–152.

Isacks, B., R. Sykes, and J. Olivier (1969), Focal me-chanisms of deep and shallow earthquakes in theTonga-Kermadec region and the tectonics of islandsarcs, Geol. Soc. Am. Bull., 80, 1443–1470.

Karig, D. E. (1971), Origin and development of mar-ginal basins in the Western Pacific, J. Geophys.Res., 76, 2542–2561.

Lagabrielle, Y., J. Goslin, H. Martin, J.-L. Thirot, andJ.-M. Auzende (1997), Multiple active spreadingcentres in the hot North Fiji Basin (SouthwestPacific): A possible model for Archaean seafloordynamics?, Earth Planet. Sci. Lett., 149, 1 –13.

Louat, R., and J. Dupont (1982), Sismicite de l’arc desTonga-Kermadec, Tech. Rep. 147, Inst. Fr. de Rech.Sci. pour le Dev. en Coop. (ORSTOM), Paris.

Lu, Z., M. Wyss, and H. Pulpan (1997), Details of stressdirections in the Alaska subduction zone from faultplane solutions, J. Geophys. Res., 102, 5385–5402.

McCaffrey, R. (1992), Oblique plate convergence, slipvector, and forearc deformation, J. Geophys. Res.,97, 8905–8915.

Millen, D., and M. Hamburger (1998), Seismologicalevidence for tearing of the Pacific plate at the north-ern termination of the Tonga subduction zone,Geology, 26, 659–662.

Nishizawa, A., N. Takahashi, S. Abe, and A. Nishizawa(1999), Crustal structure and seismicity of theHavre Trough at 26�S, Geophys. Res. Lett., 26,2549–2552.

Parson, L., and I. Wright (1996), The Lau-Havre-Taupoback-arc basin: A southward-propagating, multi-stage evolution from rifting to spreading, Tectono-physics, 263, 1 –22.

Pelletier, B., and J. Dupont (1990), Erosion, accretion,back-arc extension and slab length along the Ker-madec subduction zone, Southwest Pacific, C. R.Acad. Sci., Ser. II, 310, 1657–1664.

Pelletier, B., and R. Louat (1989), Seismotectonics andpresent-day relative plate motions in the Tonga-Lauand Kermadec-Havre region, Tectonophysics, 165,237–250.

Pelletier, B., S. Calmant, and R. Pillet (1998), Currenttectonics of the Tonga-New Hebrides region, EarthPlanet. Sci. Lett., 164, 263–276.

Pelletier, B., Y. Lagabrielle, M. Benoit, G. Cabioch,S. Calmant, E. Garel, and C. Guivel (2001), Newlyidentified segments of the Pacific-Australia plateboundary along the North Fiji transform zone, EarthPlanet. Sci. Lett., 193, 347–358.

Regnier, M. (1994), Sismotectonique de la ride de Horn(les de Futuna et Alofi), un segment en compressiondans la zone de fracture Nord-Fidjienne, C. R.Acad. Sci., Ser. II Mec., 318, 1219–1221.

Ruellan, E., and Y. Lagabrielle (2005), Oceanic subduc-tions and active spreading in the Southwest Pacific,Geomorphol. Relief ProcessusEnviron., (2), 121–142.

Ruellan, E., J. Delteil, I. Wright, and T. Matsumoto(2003), From rifting to active spreading in the LauBasin-Havre Trough backarc system (SW Pacific):Locking/unlocking induced by sesamount chain sub-duction, Geochem. Geophys. Geosyst., 4(5), 8909,doi:10.1029/2001GC000261.

Scholz, C. H., and C. Small (1997), The effect of sea-mount subduction on seismic coupling, Geology,25, 487–490.

Sdrolias, M., R. Muller, and C. Gaina (2001), Platetectonics evolution of eastern Australian marginalocean basins, in PESA Eastern Autralasian Basins

Symposium, edited by B. T. Hill, pp. 227 – 237,Petrol. Explor. Soc. of Aust., Melbourne, Victoria,Australia.

Sdrolias, M., R. Muller, and C. Gaina (2003), Tectonicevolution of the Southwest Pacific using constraintsfrom backarc basins, in Evolution and Dynamics ofthe Australian Plate, edited by R. R. Hills and R. D.Muller, Spec. Publ. Geol. Soc. Aust., 22, 343–359.

Sykes, L., B. Isacks, and J. Olivier (1969), Spatial dis-tribution of deep and shallow earthquakes of smallmagnitude in the Fiji-Tonga region, Bull. Seismol.Soc. Am., 59, 1093–1113.

Taylor, B., K. Zellmer, F. Martinez, and A. Goodliffe(1996), Sea-floor spreading in the Lau back-arc basin,Earth Planet. Sci. Lett., 144, 35–40.

Vogt, P. R., A. Lowrie, D. R. Bracey, and R. N. Hey(1976), Subduction of aseismic ridges: Effects onshape, seismicity and other characteristics of con-suming plate boundaries, Geol. Soc. Am. Spec.Pap., 172, 1 –59.

Wallace, L. M., J. Beavan, R. McCaffrey, and D. Darby(2004), Subduction zone coupling and tectonicblock rotations in the North Island, New Zealand,J. Geophys. Res., 109, B12406, doi:10.1029/2004JB003241.

Wallace, L. M., R. McCaffrey, J. Beavan, and S. Ellis(2005), Rapid microplate rotations and backarc rift-ing at the transition between collision and subduc-tion, Geology, 33, 857–860.

Wright, I. (1997), Morphology and evolution of theremnant Colville and active Kermadec Arc ridgessouth of 33�300S, Mar. Geophys. Res., 19, 177 –193.

Yu, G., S. G. Wesnousky, and G. Ekstrom (1993), Slippartitioning along major convergent plate bound-aries, Pure Appl. Geophys., 140(2), 183 –210.

Zellmer, K., andB. Taylor (2001), A three-plate kinematicsmodel for the Lau Basin opening,Geochem. Geophys.Geosyst., 2(5), doi:10.1029/2000GC000106.

��M.-A. Bonnardot, M. Regnier, E. Ruellan, and

E. Tric, Laboratoire Geosciences Azur, UMR 6526,250 rue Albert Einstein, F-06560 Valbonne, France.([email protected]; [email protected];[email protected]; [email protected])

C. Christova, Department of Seismology, Geophy-sical Institute of Bulgarian Academy of Sciences, Sofia,Bulgaria. ([email protected])

Seismicity and state of stress within the overriding plate of the Tonga-Kermadec subduction zone

Documents